Linux-audio-dev

linux-audio-dev@lists.linuxaudio.org

3 participants
7520 discussions

[linux-audio-dev] [ANNOUNCE] jack_convolve-0.0.1

by Florian Schmidt

Hi, i implemented a small convolution engine for jack.. grab it at http://affenbande.org/~tapas/jack_convolve-0.0.1.tgz untar make ./jack_convolve responsefile.wav It creates as many in/out ports as the response file has channels. Uses fftw3f and libsndfile. Will add libsamplerate support in near future. So for now, make sure the samplerate of jack and the response file match. Here's a ca 1 sec 48khz (resampled from 96khz) stereo response file of a room: http://affenbande.org/~tapas/FrontFacing%20TLM170.wav It's from this package: http://www.noisevault.com/index.php?page=3&action=file&file_id=130 which has 96khz responses.. Consumes ca. 25-30% cpu load on my 1.2ghz athlon at jack buffer size 2048 [;)] So there's plenty room for optimization (and some return value checking will be added too ;)).. If you know some tricks, let me know.. The sourcecode is pasted below for easier reference. Flo P.S.: thanks to mario lang for collaborating and giving some hints towards using fftw3f instead of fftw and some other optimizations.. P.P.S.: oh yeah, example sound, here you go [hydrogen dry then with output of jack_convolve mixed to it]: http://affenbande.org/~tapas/jack_conv_ex1.ogg And here the convoluted signal alone: http://affenbande.org/~tapas/jack_conv_ex2.ogg P.P.S.: known issues: - won't handle samplerate or buffersize changes gracefully - will bring your machine to a crawl ;) jack_convolve.cc: --------------- /* Copyright (C) 2004 Florian Schmidt This program is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details. You should have received a copy of the GNU Lesser General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. $Id: types_8h-source.html,v 1.1 2004/04/27 18:21:48 joq Exp $ */ #include <jack/jack.h> #include <iostream> #include <sstream> #include <unistd.h> #include <signal.h> #include <stdio.h> #include <sndfile.h> #include <vector> #include <cmath> #include <fftw3.h> jack_client_t *client; std::vector<jack_port_t *> iports; std::vector<jack_port_t *> oports; jack_nframes_t jack_buffer_size; int chunks_per_channel; // channel chunk data std::vector<std::vector <fftwf_complex*> > chunks; // the buffers for the fft float *fft_float; fftwf_complex *fft_complex; // the plan fftwf_plan fft_plan_forward; fftwf_plan fft_plan_backward; float normalize_factor; // per channel we need a ringbuffer holding the fft results of the // audio periods passed to us by jackd. // each needs to be sized jack_buffer_size * chunks_per_channel (see main()) std::vector<fftwf_complex *> ringbuffers; // this gets advanced by jack_buffer_size after each process() callback unsigned int ringbuffer_index = 0; // a vector to hold the jack buffer pointers.. these get resized // during init in main std::vector<jack_default_audio_sample_t *> ibuffers; std::vector<jack_default_audio_sample_t *> obuffers; // channel data std::vector<jack_default_audio_sample_t *>overlaps; int process(jack_nframes_t frames, void *arg) { // std::cout << " " << ringbuffer_index; // get pointer[s] to the buffer[s] int channels = chunks.size(); for (int channel = 0; channel < channels; ++channel) { ibuffers[channel] = ((jack_default_audio_sample_t*)jack_port_get_buffer(iports[channel], frames)); obuffers[channel] = ((jack_default_audio_sample_t*)jack_port_get_buffer(oports[channel], frames)); } for (int channel = 0; channel < channels; ++channel) { // copy input buffer to fft buffer for (int frame = 0; frame < jack_buffer_size; ++frame) { fft_float[frame] = (float)(ibuffers[channel][frame]); fft_float[frame+jack_buffer_size] = 0.0; } // fft the input[s] fftwf_execute(fft_plan_forward); // store the new result into the ringbuffer for this channel for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { ringbuffers[channel][ringbuffer_index+frame][0] = fft_complex[frame][0] / normalize_factor; ringbuffers[channel][ringbuffer_index+frame][1] = fft_complex[frame][1] / normalize_factor; } // zero our buffer for the inverse FFT, so we can simply += the // values in the next step. for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { fft_complex[frame][0] = 0; fft_complex[frame][1] = 0; } // multiply corresponding chunks of the fft'ed response[s] // we start with the chunk for the current part of the response and work our // way to the oldest data in the ringbuffer (we need to go backwards for that) for (int chunk = 0; chunk < chunks_per_channel; ++chunk) { // we go backwards and constraint to the whole buffersize ("%") long int chunk_rb_index = ((ringbuffer_index - (2 * chunk * jack_buffer_size)) + 2 * chunks_per_channel * jack_buffer_size) % (chunks_per_channel * jack_buffer_size * 2); for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { // complex multiplication (a+bi)(c+di) = (ac - bd)+(ad + bc)i long int running_ringbuffer_index = chunk_rb_index + frame; float a,b,c,d; a = ringbuffers[channel][running_ringbuffer_index][0]; b = ringbuffers[channel][running_ringbuffer_index][1]; c = chunks[channel][chunk][frame][0]; d = chunks[channel][chunk][frame][1]; fft_complex[frame][0] += (a * c) - (b * d); fft_complex[frame][1] += (a * d) + (b * c); } } // inverse fft the input[s] fftwf_execute(fft_plan_backward); // copy fft result to output buffer for (int frame = 0; frame < jack_buffer_size; ++frame) { obuffers[channel][frame] = (jack_default_audio_sample_t)(fft_float[frame] / normalize_factor); } // add previous overlap to this output buffer for (int frame = 0; frame < jack_buffer_size; ++frame) { obuffers[channel][frame] += overlaps[channel][frame]; } // save overlap for (int frame = 0; frame < jack_buffer_size; ++frame) { overlaps[channel][frame] = fft_float[frame+jack_buffer_size] / normalize_factor; } } // advance ringbuffer index ringbuffer_index += jack_buffer_size * 2; ringbuffer_index %= jack_buffer_size * 2 * chunks_per_channel; return 0; } bool quit = false; void signalled(int sig) { std::cout << "exiting.." << std::endl; quit = true; } int main(int argc, char *argv[]) { // we need to become jack client first so we can ask for the buffer // size. std::cout << "jack_convolve (C) 2004 Florian Schmidt - protected by GPL2" << std::endl; if (argc < 2) { std::cout << "usage: jack_convolve responsefile.wav" << std::endl; exit(0); } // hook up signal handler for ctrl-c signal(SIGINT, signalled); client = jack_client_new("convolve"); jack_buffer_size = jack_get_buffer_size(client); normalize_factor = sqrt(2.0 * (float)jack_buffer_size); std::cout << "buffer size: " << jack_buffer_size << std::endl; // first we load the response file. we simply assume it has // the right samplerate ;) the channel count of the // response file governs how many Ins and Outs // we provide to jack.. // filename of the soundfile is the first commandline // parameter, argv[1] struct SF_INFO sf_info; SNDFILE *response_file = sf_open (argv[1], SFM_READ, &sf_info) ; // register ports for each channel in the response file std::cout << "channels in response file: " << sf_info.channels << std::endl; std::cout << "registering ports:"; for (int i = 0; i < sf_info.channels; ++i) { std::stringstream stream_in; std::stringstream stream_out; stream_in << "in" << i; stream_out << "out" << i; std::cout << " " << stream_in.str(); std::cout << " " << stream_out.str(); jack_port_t *tmp_in = jack_port_register(client, stream_in.str().c_str(), JACK_DEFAULT_AUDIO_TYPE, JackPortIsInput, 0); jack_port_t *tmp_out = jack_port_register(client, stream_out.str().c_str(), JACK_DEFAULT_AUDIO_TYPE, JackPortIsOutput, 0); iports.push_back(tmp_in); oports.push_back(tmp_out); } std::cout << std::endl; std::cout << "length of response file in frames: " << sf_info.frames << std::endl; if (sf_info.samplerate != jack_get_sample_rate(client)) { std::cout << "warning: samplerate in responseFile: " << sf_info.samplerate << "; jack-samplerate: " << jack_get_sample_rate(client) << std::endl; std::cout << "will resample response file" << std::endl; } // find out how many chunks we need per channel: chunks_per_channel = (int)ceil((float)sf_info.frames/(float)jack_buffer_size); std::cout << "chunks per channel: " << chunks_per_channel << std::endl; // allocate chunk memory for (int i = 0; i < sf_info.channels; ++i) { std::vector<fftwf_complex*> channel; for (int j = 0; j < chunks_per_channel; ++j) { // zero padded to twice the length fftwf_complex *tmp = (fftwf_complex*)fftwf_malloc(sizeof(fftwf_complex) * jack_buffer_size * 2); // zero for (int frame = 0; frame < jack_buffer_size * 2; ++frame) tmp[frame][0] = tmp[frame][1] = 0; channel.push_back(tmp); } chunks.push_back(channel); } std::cout << "chopping response file..."; // fill the chunks with the appropriate data float *tmp = new float[sf_info.channels]; for (int chunk = 0; chunk < chunks_per_channel; ++chunk) { for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { // pad with 0's if (chunk*jack_buffer_size + frame < sf_info.frames && frame < jack_buffer_size) { int result = sf_readf_float(response_file, tmp, 1); if (result != 1) std::cout << "problem reading the soundfile" << std::endl; } else { for (int channel = 0; channel < sf_info.channels; ++channel) { tmp[channel] = 0; } } for (int channel = 0; channel < sf_info.channels; ++channel) { // set real value to sound data chunks[channel][chunk][frame][0] = tmp[channel]; // std::cout << tmp[channel] << " "; // imaginary value to 0 chunks[channel][chunk][frame][1] = 0; } } } std::cout << "done." << std::endl; std::cout << "creating fftw3 plan..."; // ok, now we need to FFT each chunk.. For this we need an FFT plan. // buffers fft_float = new float[jack_buffer_size * 2]; fft_complex = (fftwf_complex*)fftwf_malloc(sizeof(fftwf_complex) * jack_buffer_size * 2); // create fftw plan fft_plan_forward = fftwf_plan_dft_r2c_1d(jack_buffer_size * 2, fft_float, fft_complex, FFTW_MEASURE); fft_plan_backward = fftwf_plan_dft_c2r_1d(jack_buffer_size * 2, fft_complex, fft_float, FFTW_MEASURE); std::cout << "done" << std::endl; // fft the chunks std::cout << "FFT'ing response file chunks..." << std::endl; for (int channel = 0; channel < sf_info.channels; ++channel) { std::cout << "channel: " << channel << ": "; for (int chunk = 0; chunk < chunks_per_channel; ++chunk) { std::cout << "."; // copy chunk to input buffer for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { fft_float[frame] = chunks[channel][chunk][frame][0]; // fft_in[frame][1] = 0; } // fft fftwf_execute(fft_plan_forward); // copy output buffer to chunk for (int frame = 0; frame < jack_buffer_size * 2; ++frame) { chunks[channel][chunk][frame][0] = fft_complex[frame][0] / normalize_factor; chunks[channel][chunk][frame][1] = fft_complex[frame][1] / normalize_factor;; } } std::cout << std::endl; } std::cout << "done." << std::endl; // make room so we can store the buffer pointers for each channel ibuffers.resize(sf_info.channels); obuffers.resize(sf_info.channels); // allocate ram for ringbuffers and zero out for 0 noise :) for (int channel = 0; channel < sf_info.channels; ++channel) { fftwf_complex *tmp = (fftwf_complex*)fftwf_malloc(sizeof(fftwf_complex) * jack_buffer_size * 2 * chunks_per_channel); // zero out buffers for (int frame = 0; frame < jack_buffer_size * chunks_per_channel * 2; ++frame) { tmp[frame][0] = 0; tmp[frame][1] = 0; } ringbuffers.push_back(tmp); } // allocate buffers for overlap for (int channel = 0; channel < sf_info.channels; ++channel) { jack_default_audio_sample_t *tmp = new jack_default_audio_sample_t[jack_buffer_size]; overlaps.push_back(tmp); } // now we should be ready to go // std::cout << chunks.size() << std::endl; jack_set_process_callback(client, process, 0); jack_activate(client); std::cout << "running (press ctrl-c to quit)..." << std::endl; while(!quit) {sleep(1);}; // sleep(seconds_to_run); jack_deactivate(client); jack_client_close(client); } --------------- -- Palimm Palimm! http://affenbande.org/~tapas/

20 years, 10 months

[linux-audio-dev] Tuning

by james＠dis-dot-dat.net

Back in the days before I converted my windows partition to ext3, I used Cubase with a number of VST instruments and effects. I miss a number of these, but there is usually something equivalent or something that will one day be equivalent in the Linux world. One thing I haven't been able to replace so far is the Oberheim OB-Tune plug-in. This was an amazingly useful plug-in that would take an audio input and make sure it stayed in tune. It worked on guitars, vocals, synths, whatever. There were a number of ways to use it. You could define a set of notes that were allowed, and audio would be made to be in tune with the closest note (useful for phrases) or you could specify a single note. This was all parameterised, so you could write changes to these throughout a track. Is there anything like this out there at the moment for Linux? If there isn't, does anyone have any idea how it works? I might consider implementing something similar, but I'm not sure where to begin. Here's what I'm thinking: Operate in smallish chunks. Find the most intense frequency (FFT or such) and decide how far that is from the desired frequency. Scale accordingly, preferably with as little distortion as possible, so pack and crossfade sections. If anyone has any thoughts, please let me know. I'd expect to make this a plug-in rather than a standalone app, but I've never touched LADSPA before - is it possible to send events to change parameters? This might be more than I can chew, but I intend to bite anyway. James -- "I'd crawl over an acre of 'Visual This++' and 'Integrated Development That' to get to gcc, Emacs, and gdb. Thank you." (By Vance Petree, Virginia Power)

20 years, 10 months

[linux-audio-dev] test -- please ignore

by NadaSpam

Just trying to find out what's so "suspicious" about my header.

20 years, 10 months

[linux-audio-dev] lad design patterns

by Levi D. Burton

does the idea of documenting various lad design patterns make sense to anyone? seems a lot of stuff is sort of voodooish. example: application timer design or, am i just a moron and should be able to come up with this sort of stuff easily and on my own, in other words: common sense? thanks. -- Levi D. Burton http://www.puresimplicity.net/~ldb/

20 years, 10 months

[linux-audio-dev] Re: [Jackit-devel] Re: [linux-audio-user] [ANNOUNCE] jack_convolve-0.0.1

by Florian Schmidt

On Fri, 28 Jan 2005 21:26:46 -0800 (PST) R Parker <rtp405(a)yahoo.com> wrote: > Florian, > > Just gave your dry and wet sample a quick listen. It > sounds promising. I'm looking forward to giving it a > more serious run. > Hi, make sure you get the newest version.. homepage here: http://www.affenbande.org/~tapas/wiki/admin.php?jack_convolve tarball[s] here [minor optimization attempts since last tarball]: http://affenbande.org/~tapas/jack_convolve/ I also created another example file where i use a different response file [the one used for the other two examples was noisy]. hear you hear the convolution result get mixed in and out: http://affenbande.org/~tapas/jack_conv_ex3.ogg Flo -- Palimm Palimm! http://affenbande.org/~tapas/

20 years, 10 months

[linux-audio-dev] Buggy audio debian packages

by Eric Dantan Rzewnicki

Hi, I just saw the latest list of debian package bugs and noticed the packages for these apps listed: Package: icecast-server (debian/main) Maintainer: Emmanuel le Chevoir <mms(a)debian.org> 292233 [ ] manpage looks not distributeable, as license not adhered Package: specimen (debian/main) Maintainer: Eduardo Marcel Macan <macan(a)debian.org> 268317 [ + ] [UX] specimen: FTBFS on amd64: cast to pointer from integer of different size 291252 [ ] [X] specimen: Uninstallable on sid - library dependency does not exist. Package: supercollider (debian/main) Maintainer: Paul Brossier <piem(a)altern.org> 290339 [ H ] [X] supercollider: [m68k s390 mips sparc arm] FTBFS Package: swami (debian/main) Maintainer: Guenter Geiger (Debian/GNU) <geiger(a)debian.org> 279154 [ ] [EX] swami files are in /debian/.... Is anyone here working on fixing any of these? or know someone who is? -- Eric Dantan Rzewnicki | Systems Engineer I Technical Operations Division | Radio Free Asia 2025 M Street, NW | Washington, DC 20036 | 202-530-4900 CONFIDENTIAL COMMUNICATION This e-mail message is intended only for the use of the addressee and may contain information that is privileged and confidential. Any unauthorized dissemination, distribution, or copying is strictly prohibited. If you receive this transmission in error, please contact network(a)rfa.org.

20 years, 10 months

[linux-audio-dev] [ANN] Aqualung 0.9beta4 released

by Tom Szilagyi

Aqualung 0.9beta4 released http://aqualung.sf.net I am pleased to announce the fourth public beta release of Aqualung, a music player for GNU/Linux. Aqualung plays audio files from your filesystem and has the feature of sample-accurate track changing (no gaps inserted between adjacent tracks). It also supports high quality sample rate conversion between the file and the output device, when necessary. Supported formats include: almost all sample-based, uncompressed files (eg. WAV, AIFF, AU etc.), FLAC (the Free Lossless Audio Codec), Ogg Vorbis, MPEG Audio (including the infamous MP3 format) and MOD audio formats (MOD, S3M, XM, IT, etc.). The program can play the music through OSS, ALSA or the JACK Audio Connection Kit. Other features include LADSPA plugin support, multiple skin support & state persistence via XML config files. The so-called Music Store provided by Aqualung is an XML-based music database for arranging your music collection. This new release adds many new features, including file metadata (FLAC/Vorbis/ID3) display & importing, volume calculation and playback RVA (relative volume adjustment) support. All users are encouraged to upgrade. Download the new version from the project homepage: http://aqualung.sf.net You can read the changelog below. Hope you enjoy this release. Tom * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * 2005-01-28 Tom Szilagyi <tszilagyi at users dot sourceforge dot net> * Aqualung 0.9beta4 http://aqualung.sf.net INCOMPATIBLE CHANGES: [none] NEW LIBRARY DEPENDENCIES: * liblrdf 0.4.0 is now required (was: 0.3.7) http://lrdf.sourceforge.net * libid3tag library required if you want ID3v2 support http://sourceforge.net/project/showfiles.php?group_id=12349 MAJOR CHANGES: * Internationalization support via gettext. German, Hungarian and Ukrainian translations available; new translations for any language happily accepted at any time * Implemented read support for .m3u and .pls playlist formats. The formats are distinguished using file extensions (case insensitive). Now you can supply .m3u or .pls files on the command line, or select one in the Load/Enqueue Playlist option of the playlist window's popup menu. Aqualung does not implement shoutcast ATM, so URLs will be simply discarded. * new 'File info' dialog box (accessible from the Music Store and the Playlist) displays FLAC stream metadata, Ogg Vorbis comments and ID3v2 tags found in the soundfiles. * playback RVA support. Aqualung has its own system for this, from volume level calculation of files in the Music Store, to adjusting the dynamics characteristics to your listening environment. * Import FLAC/Vorbis/ID3v2 metadata into the Music Store via the 'File info' dialog accessed from the Music Store. On the right side of the tag data fields, there are buttons to import every piece of information into relevant fields of the Music Store database. In particular, ID3v2.4 RVA tags can be imported as manual RVA adjustment values. * many changes to enable displaying track lengths and RVA values in the Playlist. You can configure the column order in the Playlist, and displaying Lengths and RVA values can be turned off. (Track lengths are shown on the right side, RVA is hidden by default.) * New Settings notebook page "Playlist" for configuring the behavior of this stuff. * new remote option to terminate an already running instance: the -Q or --quit option will cause the instance specified by -N, or the 0-th instance by default, to terminate (just as if you exited it normally). * added support for remotely changing the volume via the --volume or -V option. Defaults to the 0th running instance. * major code rearrangement of Aqualung Core (file decoder is abstracted, runnable in multiple instances at the same time, separated in file_decoder.[ch]) * added all four basic aqualung skins (dark, default, metal, woody) to CVS. These are automatically available after a make install from now, no need to install them separately. * also, the skins have been updated to match recent new dialogs & widgets * docs update (manpage, HTML) for the beta4 release. MINOR CHANGES: * ./configure won't stop anymore if an optional library is missing, unless --with-PACKAGE is applied. Closes mantis bug #16. * Changes to the interface for adding files to the playlist locally or remotely * Implemented conversion from/to UTF8/locale charset. (Closes mantis bug #7). Note: please set the environment variable G_BROKEN_FILENAMES or G_FILENAME_ENCODING appropriately if your filesystem encoding is not UTF8. See http://developer.gimp.org/api/2.0/glib/glib-running.html for details. * Starting playback of a new track via double-clicking on it in the playlist is now allowed when another track is being paused. (Closes bug #14) * Implemented cue-from-paused-state functions (re: bug #15) as suggested by SGh. * Added check for pkg-config -exist "jack" to configure.ac. Up till now, the ./configure script failed to detect the condition when the jack daemon is present, but development files are not. (As if you installed JACK from a distro, but forgot to include the corresponding -dev package.) * Implement trashlist object to collect and eventually free pieces of memory that are malloc'ed in an ad-hoc manner in different places, but need to be freed sometime. * Use the trashlist to collect and free some memory that was leaked until now in LADSPA plugin dialogs and the File info dialog. * renamed "Options" dialog to "Settings". This name better suits the purpose of this dialog. * minor rearrangement of "Add Artist", "Edit Artist", "Add Record", "Edit Record", "Add Track" & "Edit Track" dialogs, hopefully for the better. * changed About box font to normal (Courier is not always available) * Workaround LADSPA plugin loading bug on ReiserFS (use the --with-brokenplfix configure option)

20 years, 10 months

[linux-audio-dev] Re: Tuning

by Stefan Turner

The other suggestions should good, but alternatively you write a patch in Pure Data to do this, using the built-in object fiddle~, which estimates pitch (fairly accurate with tweaking) and a LADSPA pitch shifter (the best I've found is pitchScaleHQ from the collection at www.plugin.org.uk). PD takes a while to learn how to use, but once you have it's great for doing stuff like this quickly. Stefan Turner ___________________________________________________________ ALL-NEW Yahoo! Messenger - all new features - even more fun! http://uk.messenger.yahoo.com

20 years, 10 months

[linux-audio-dev] Re: Tuning

by Stefan Turner

>Thanks for the advice. > >I've never used pd - can I use it to make a LADSPA >plugin or >standalone app, No, unfortunately not. PD 'interprets' the patch, but it can use Jack and run LADSPA plugins inside it... though once you get into it the ability to completely customise effects, synths etc. often makes the idea of plugins seem a little restricted. It's well documented and there's a good mailing list for it, so post there if you have any problems. > or do I have to run it managed? Sorry I don't know what this means, so I guess the answer's no. Good luck Stefan Turner ___________________________________________________________ ALL-NEW Yahoo! Messenger - all new features - even more fun! http://uk.messenger.yahoo.com

20 years, 10 months

[linux-audio-dev] M-Audio Transit USB with ALSA?

by Dr.Graef＠t-online.de

Hi, has anyone tried the M-Audio Transit USB with ALSA yet? (Its bigger cousin, the USB Audiophile is supported according to alsa-project.org.) I'm looking for a not-too-expensive USB audio interface for my laptop, and the Transit would fit nicely, as I already have a usb midi interface. Albert -- Dr. Albert Gr"af Dept. of Music-Informatics, University of Mainz, Germany Email: Dr.Graef(a)t-online.de, ag(a)muwiinfa.geschichte.uni-mainz.de WWW: http://www.musikwissenschaft.uni-mainz.de/~ag

20 years, 10 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Linux-audio-dev